Project 2 Report: Natural disaster

University of Sydney | DATA1901 | Oct 2024

Author

SID 540786875

Client Bio and Recommendation

Client:

Niels Holm-Nielsen - Head of the Global Facility for Disaster Reduction and Recovery (GFDRR). (LinkedIn|Website)

GFDRR, administered by the World Bank Group, supports low-income countries on disaster risk mititgation. Since 2015, GFDRR has mobilized around $35 billion in financing for disaster and climate resilience operations worldwide, making a significant impact on reducing vulnerability to natural hazards.

Holm-Nielsen joined the World Bank in 2006 and has held key positions in Disaster Risk Management across regions such as Latin America, the Caribbean, and Africa, most recently serving as Lead Disaster Risk Management Specialist and Global Lead for Resilience and Disaster Risk Management.

Recommendation:

It is recommended that Holm-Nielsen significantly increase funding for disaster risk reduction in Southern and Eastern Asia. The regression analysis demonstrates a link between economic damage and progress in risk reduction, suggesting that enhanced efforts can reduce economic losses. Allocating more resources to countries like Nepal, Vietnam, Cambodia, Thailand, Pakistan, and the Philippines will be a strategic investment in sustainable development that improves economic resilience.

Evidence

The analysis focuses on three variables: 

  • Number of people affected by disasters per 100,000 and Total economic damages from disasters as a share of GDP from the Natural Disaster data. 
  • Risk reduction progress score from Gapminder’s Natural risk reduction progress score data.
Code
# Load libraries
library(tidyverse)
library(ggplot2)
library(plotly)
library(maps)
library(viridis)
library(rnaturalearth)
library(rnaturalearthdata)

# Read and clean datasets
disaster_data <- read.csv("natural-disasters.csv")

disaster_data <- disaster_data %>% 
  rename(EconDmgShare = Total.economic.damages.from.disasters.as.a.share.of.GDP, 
         HumanDmgPer = Number.of.people.affected.by.disasters.per.100.000) %>%
  filter(!is.na(EconDmgShare),
         !is.na(HumanDmgPer),
         !Entity %in% c("High income", "Low income", "Lower middle income", "Upper middle income", "World"),
         Year >= 1960) %>%
  select(Entity, Year, EconDmgShare, HumanDmgPer)

score_data <- read.csv("risk-reduction-score.csv")

score_data <- score_data %>% 
  rename(RiskReductionScore = X2011)

I. Human and Economic Vulnerability in Asia

Code
regions_list <- c("Africa", "Asia", "North America", "South America", "Europe", "Oceania")

regional_data <- disaster_data %>%
  filter(Entity %in% regions_list)

regional_impact <- regional_data %>% 
  group_by(Entity) %>%
  summarise(across(c(`EconDmgShare`, `HumanDmgPer`), sum, na.rm = TRUE))

plot1 <- ggplot(regional_impact, aes(x = Entity, y = `HumanDmgPer`, fill = Entity)) +
  geom_bar(stat = "identity", position = "dodge", width = 0.75) +
  labs(title = "Regional Sociological Impact of Natural Disasters (1960 - 2010)", y = "People affected per 100,000", x = "Region") +
  theme_minimal() +
  scale_fill_viridis_d(option = "mako") +
  theme(legend.position = "none")

ggplotly(plot1)
Code
plot2 <- ggplot(regional_impact, aes(x = Entity, y = `EconDmgShare`, fill = Entity)) +
  geom_bar(stat = "identity", position = "dodge", width = 0.75) +
  labs(title = "Regional Economic Damage of Natural Disasters (1960 - 2010)", y = "Economic damage as a share of GDP (%)", x = "Region") +
  theme_minimal() + 
  scale_fill_viridis_d(option = "mako") +
  theme(legend.position = "none")


ggplotly(plot2)

The bar plots reveal a high disparity in human damage between Asia and the rest of the world. While ranking second to Oceania in economic damage, it is also significant, highlighting Asia’s overwhelming vulnerability to natural disaster and need of damage mitigation and recovery support.

II. High-Risk Zones in Southern and Eastern Asia

Code
# Filter data
dmg2010_data <- disaster_data %>%
  filter(Year == 2010, 
         !is.na(EconDmgShare)) %>%
  select(Entity, EconDmgShare)

map_data <- ne_countries(scale = "medium", returnclass = "sf", continent = "asia")

# Merge the cleaned disaster data with the world map data
map_data <- map_data %>%
  left_join(dmg2010_data, by = c("name" = "Entity"))

# Map
my_map <- ggplot(data = map_data) +
  geom_sf(lwd = 0.2, 
          aes(fill = EconDmgShare,
              text = paste("Country:", name, "<br>Economic Damage:", round(EconDmgShare, 2)))) +
  scale_fill_viridis(option = "viridis", 
                     na.value = "grey50", 
                     breaks = seq(0, 2.402, by = 0.4)) +
  labs(title = "Total Economic Damages from Disasters as a Share of GDP by Country in 2010",
       fill = "Economic\nDmg (%)",
       x = "Longitude", 
       y = "Latitude") +
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5))

ggplotly(my_map, tooltip = "text")

The interactive map, visualizing economic damage across Asian countries, indicates that GFDRR should prioritize resources for countries such as Nepal, Vietnam, Cambodia, Thailand, Pakistan, and the Philippines. This aligns with findings from the ASEAN Disaster Risk Management Initiative (UNDRR 2010), which identifies Southeast Asia as one of the most disaster-prone regions in the world. Given that all these countries are located in Southern and Eastern Asia, there is a strong case for focusing disaster mitigation effort in this vulnerable area.

III. The Case for Investing in Resilience

Code
sea_countries_list <- c("Bangladesh", "Brunei", "Cambodia", "China", "Fiji", "India", "Indonesia", "Japan", "Kiribati", "Laos", "Malaysia", "Myanmar", "Nepal", "North Korea", "Pakistan", "Palau", "Philippines", "Singapore", "South Korea", "Sri Lanka", "Thailand", "Timor", "Tuvalu", "Vietnam")

# Join with score data and filter
final_data <- score_data %>%
  left_join(dmg2010_data, by = c("country" = "Entity")) %>%
  filter(country %in% sea_countries_list) %>%
  na.omit()

plot3 <- ggplot(final_data, aes(x = EconDmgShare, y = RiskReductionScore)) +
  geom_point(aes(text = paste("Country: ", country, 
                              "<br>Score: ", RiskReductionScore, 
                              "<br>Economic Damage: ", round(EconDmgShare, 2)))) +
  geom_smooth(method = "lm", se = FALSE, color = "#42f5b3") +
  labs(x = "Economic Damage as a share of GDP (2010)",
       y = "Risk Reduction Progress Score (2009 - 2011)",
       title = "Correlation between Risk Reduction Progress Score and Economic Damage") +
  theme_minimal()

ggplotly(plot3, tooltip = "text")
Code
correlation <- round(cor(final_data$RiskReductionScore, final_data$EconDmgShare), 2)
print(paste("Correlation: ", correlation))
[1] "Correlation:  -0.59"

The disaster risk reduction progress score is based on self-assessment scores submitted by countries under the Hyogo Framework (UNDRR) from 2009 to 2011, ranging from 1 to 5. The regression model indicates a moderate negative correlation between economic damage as a share of GDP in 2010 and the score for Southern and Eastern Asian countries. In line with the findings of the Asian Development Bank, which states that effective disaster risk reduction can yield $4–$7 in savings for every dollar invested, supporting the argument for GFDRR to increase funding in these regions

IV. Hypothesis Testing

To determine if economic damages of countries in the Southern and Eastern Asia area is significantly higher than the rest of Asia, a proportion test is utilized.

  • p-value = 0.02243 < 0.05

The p-value falls below the threshold of 0.05, supporting the conclusion that a greater number of countries in Southern and Eastern Asia experience substantial economic damage.

Appendix

Niels Holm-Nielsen is a solid choice due to his leading position at GFDRR and extensive experience in disaster risk management. Since 2015, GFDRR has mobilized about $35 billion for disaster and climate resilience initiatives. Holm-Nielsen’s leadership has been crucial to this important work of reducing vulnerability to natural disasters. This report provides Holm-Nielsen with insight on which region the organization should focus on, and how enhancing risk mitigation might benefit a country’s economic growth.

1. Hypotheses

Since around half of Asian countries are in the Southern and Eastern Asia, we expect half of the countries that experienced high economic damage (>0.5% of GDP) to be in the region.

H0: 50% of Asian countries that experienced high economic damage from natural disaster are from Southern and Eastern Asia.

H1: More than 50% of Asian countries that experienced high economic damage from natural disaster are from Southern and Eastern Asia.

2. Assumptions

  • Database Accuracy: All countries’ statistics are correct.
  • Independence: Countries’ damage should be independent of each other.

3. Test Statistics

Code
asia_countries_list <- c("Afghanistan", "Armenia", "Azerbaijan", "Bahrain", "Bangladesh", "Bhutan", "Brunei", "Cambodia", "China", "Cyprus", "Georgia", "India", "Indonesia", "Iran", "Iraq", "Israel", "Japan", "Jordan", "Kazakhstan", "Kuwait", "Kyrgyzstan", "Laos", "Lebanon", "Malaysia", "Maldives", "Mongolia", "Myanmar", "Nepal", "North Korea", "Oman", "Pakistan", "Palestine", "Philippines", "Qatar", "Russia", "Saudi Arabia", "Singapore", "South Korea", "Sri Lanka", "Syria", "Taiwan", "Tajikistan", "Thailand", "Timor-Leste", "Turkey", "Turkmenistan", "United Arab Emirates", "Uzbekistan", "Vietnam", "Yemen")

asia_damage_data <- disaster_data %>%
  filter(Entity %in% asia_countries_list) %>%
  select(Entity, Year, EconDmgShare)

high_damage_data <- asia_damage_data %>% filter(EconDmgShare > 0.5)

sea_count <- high_damage_data %>% filter(Entity %in% sea_countries_list) %>% nrow()
total_count <- nrow(high_damage_data)

test_stat <- (sea_count - 0.5 * total_count) / sqrt(0.5 * (1 - 0.5))
print(paste("Test statistics:", test_stat))
[1] "Test statistics: 14"

4. P-value

Code
prop.test(sea_count, total_count, 0.5, alternative = "greater")

    1-sample proportions test with continuity correction

data:  sea_count out of total_count, null probability 0.5
X-squared = 4.0238, df = 1, p-value = 0.02243
alternative hypothesis: true p is greater than 0.5
95 percent confidence interval:
 0.5280587 1.0000000
sample estimates:
        p 
0.6666667 

5. Conclusion

Reject H0

  • Data on natural disasters is limited, with only one data point every ten years.
  • The Natural Disaster database includes records only up to 2010.
  • Certain countries and territories are unable to provide data.

Ethics Statement

  • Shared Professional Values: The project follows the shared value of Respect by acknowledging and properly crediting the works of others, ensuring that their work was neither suppressed nor diminished.

  • Ethical Principles: The project upheld the principle of Bearing Responsibility for the Integrity of the Discipline by ensuring transparency in methods and findings, avoiding any form of deception or misrepresentation.

Acknowledgements

Name Type of Resource URL/Source
Topic 2 & 8 Extensions Ed Lessons https://edstem.org/au/courses/16787/lessons/
Introduction to the viridis color maps Documentation https://cran.r-project.org/web/packages/viridis/vignettes/intro-to-viridis.html
How to remove border line when using geom_sf() Blog/Post https://stackoverflow.com/questions/47720329/how-can-i-remove-border-lines-when-using-geom-sf
Grammarly Spell/Grammar Checker https://app.grammarly.com
ChatGPT Generative AI https://app.grammarly.com

References

The Asian Development Bank - Synthesis Report on Ten ASEAN Countries Disaster Risks Assessment, https://www.unisdr.org/files/18872_asean.pdf

UNDRR: Financing Disaster Risk Reduction in Asia and the Pacific: A Guide for Policy Makers, https://dx.doi.org/10.22617/TIM200415-2

UNDRR: Disaster risk reduction progress score and Hyogo Framework for Action, https://www.preventionweb.net/sendai-framework/Hyogo-Framework-for-Action